TCA9539-Q1: CLK freq and tsch spec abnormal

Scarlett Cao

Part Number: TCA9539-Q1

Hi team,

Customer is evaluating TCA9539-Q1 and found the output freq is abnormal and highly dependent on pull up resistor, as below, would you pls share if the possible reason?

When I2C configure to 400kHz

1. I2C pull up 4.64k, measured CLK freq is 377.385kHz, tsch I2C clock high time is only 516ns, lower than spec

2. I2C pull up 1.5k, measured CLK freq is 433kHz

3. I2C pull up 1k, measured CLK freq is 452kHz

ThX

Scarlett

over 1 year ago

0 Jack Guan over 1 year ago

TI__Mastermind 40526 points

Hey Scarlett,

I2C data rate is controlled entirely by the external pullups + capacitance on the bus. You may see this app note on how to select the proper pullup depending on the I2C speed you need: I2C Bus Pull-Up Resistor Calculation

Thanks,

Jack

0 Scarlett Cao over 1 year ago in reply to Jack Guan

TI__Expert 4000 points

Hi Jack

According to the file, R and Cb influence the rise time, it didnt say how freq may depend on Rp. Could you help support analyze the pull up pull down influence to frequency?

Thanks

Scarlett

0 Tyler Townsend over 1 year ago in reply to Scarlett Cao

TI__Mastermind 47668 points

Hi Scarlett,

The change of R and Cb shouldn't change the frequency of the signal but only the rise-time. How is the customer measuring the i2c clock frequency and do we have supporting O-scope shots to support the case?

Do we have any other devices on the bus that are between the MCU and the TCA9539-Q1?

Switches, buffers, level-shifters etc?

Regards,

Tyler

0 Clemens Ladisch over 1 year ago in reply to Tyler Townsend

Guru 316880 points

I²C masters must monitor the bus signals to detect arbitration failures or clock stretching. When a master wants to output a rising edge on SCL, it ceases to pull the line down, and then waits for the voltage to actually rise. This waiting time depends both on the actual rise time and on the master implementation.

The TCA9539-Q1 never does clock stretching, so it cannot affect the frequency.

0 Tyler Townsend over 1 year ago

TI__Mastermind 47668 points

Hi Scarlett,

Please let me know if there are any comments to my questions or Clemens questions about clock stretching or multi-master arbitration.

Regards,

Tyler

Interface

Interface forum

TCA9539-Q1: CLK freq and tsch spec abnormal